¹ All page references in the text are to Barry, B., Justice as Impartiality, Oxford, 1995Google Scholar. It should be noted that this book is the second volume of what is now conceived as a tetralogy, of which the third and fourth volumes have yet to appear. A degree of tenta-tiveness on some issues is therefore appropriate; but as Barry himself says the present book ‘is designed to be self-contained’ (p. ix) it is not unreasonable to treat it as such. I discussed the first volume, Theories of Justice, London, 1989Google Scholar, in ‘A Theory of Social Justice?’, Utilitas, iii (1991), 121–38Google Scholar.

² It may seem that my strategy in this paper is wholly unsuited to establishing this larger claim, comprising as it does of arguments that often take up highly specific features of Barry's theory. However, I am inclined to think that all such theories can only be addressed on a case by case basis; for there probably is no conclusive knock-down argument which refutes them all. The larger conclusion about the nature of the project of formulating a general theory of justice, therefore, arises from seeing that all attempts to undertake it so far share similar failings, and concluding that this gives us reason to think that none are likely to succeed.

³ Although it should be strictly irrelevant, it may be worth mentioning that I share many of Barry's substantive views about the scope of law and the desirability of economic redistribution. What I dissent from is his belief that justice as impartiality provides a justification for them that cannot reasonably be rejected.

⁴ In view of the extraordinary length at which Barry is presenting his theory – 711 pages with two volumes still to come – this may seem a surprising claim. However, the length of the two books so far published is due in no small measure to Barry's apparent reluctance to resist exploring any argumentative byway which catches his eye. While many of these detours have their own interest, I suspect I am not the only reader to wish that in some areas more attention had been given to explaining in greater detail the crucial moves in his own theory, even at the cost of sacrificing some of the undeniably interesting but less central material.

⁵ My order of exposition does not follow Barry's but aims to be a reasonably faithful representation of his line of thought. Given its brevity, however, it cannot hope to present the structure of his argument in anything other than the most skeletal form.

⁶ The particular idea of principles ‘which no one could reasonably reject’, as indeed the general specification of the contract situation, comes from Scanlon, T. M., ‘Contractualism and Utilitarianism’, Utilitarianism and Beyond, ed. Sen, A. and Williams, B., Cambridge, 1992, pp. 103–8Google Scholar. Three points of difference between Barry and Scanlon are, however, particularly worth noting. First, Scanlon is concerned with a theory of morality, Barry with the more limited aim of a theory of justice. Second, Scanlon believes there is a significant difference between principles ‘which everyone could reasonably reject’ and principles ‘which everyone could reasonably accept’ by contrast Barry thinks nothing of significance hangs on this difference and uses them interchangeably. These two differences are both acknowledged by Barry himself. The third, however, is not. Scanlon's approach is that ‘reasonable rejection’ has nothing to do with the number of people who would object. Yet Barry, as I shall show later, frequently employs quantitative adjectives to qualify the scope of his arguments in a way which threatens to under-mine the entire logic of the Scanlonian contract.

⁷ An alternative approach, adopted by Rawls, John in his Political Liberalism, New York, 1993Google Scholar, is to employ a strategy of avoidance, a strategy which attempts to escape entering into any epistemological arguments.

⁸ It may be unfair to expect such a list but it would at least be helpful if Barry set out some criteria by which these considerations are to be identified. Presumably we are supposed to arrive at principles which would be reasonable through what Barry labels the ‘a priori’ or ‘empirical’ methods (§§ 32–4). However, it remains unclear, to this reader at least, how substantive conclusions do arise. We must hope that the next volume contains some more examples worked out in detail of how particular substantive principles are supposed to be derived from the theoretical structure of justice as impartiality.

⁹ There is a good reason to believe that if the poll tax had been set at a much lower level, as it could have been, then opposition to it would have been significantly less.

¹⁰ Rawls, John, A Theory of Justice, Oxford, 1971, § 60Google Scholar.

¹¹ Particularly worrying is the way in which moral principles, concern for others and self-interest are all lumped together. It is true that Barry acknowledges that ‘conceptions of the good are complex systems of belief’ and that ‘we must resist any temptation … to treat them as “preferences”, of the same underlying nature as a taste for strawberry ice-cream’ (p. 167). But this acknowledgement does not seem to have implications for his theory.

¹² For example, Barry, writes: ‘Clearly I have introduced substantive moral ideas in the course of talking about what could be reasonably rejected. Since I have already said that nothing can be expected from the bare notion of rationality itself, I am not in the least embarrassed by recognizing that this is so’ (p. 8)Google Scholar. Leaving aside the question of his propensity to be embarrassed, what eludes me is the explanation of why the moral ideas to which Barry helps himself are (uniquely?) thought to float free of any conception of the good.

¹³ There should be a convergence between the principles derived from the theory and what people will in fact agree to under conditions of equality of power and full information. However, given that these conditions are rarely to be found in our world, actual agreements have to be treated with considerable suspicion; they are much more likely to be based on power differentials or incomplete information. The empirical conditions in the world that tend to promote just outcomes are, in Barry's terminology, ‘the circumstances of impartiality’.

¹⁴ In a review of Theories of Justice, Will Kymlicka raised an important problem for Barry: ‘It is understandable (although I think false) to say that my desire to act morally towards other competent adults is tied to a desire to justify my actions to them. But what does it mean to desire impartial agreement with infants or to be able to justify one's actions to people who don't yet exist? … The emphasis on agreement within impartiality seems to create some of the same problems that the emphasis on bargaining power creates within mutual advantage theories: some people will fall beyond the pale of morality, including those who are most in need of protection’ (Kymlicka, W., ‘Two Theories of Justice’, Inquiry, xxxiii (1990), 110)Google Scholar. It is fairly clear from the present volume that agreement does little, if any, real work in Barry's theory – as is often the case in contractarian theories there is an appearance of voluntarism but the appearances are deceptive.

¹⁵ In Theories of Justice Barry made great play of the international and intergenerational aspects of his theory. These are much less evident in Justice as Impartiality. In the former book Barry really did seem to want to include as parties to the agreement everyone living and who would live. Barry does not retract this in the present book but nor does he make anything of it. There are in any case especially formidable problems about the representation of future generations within a contractarian model, not least because principles adopted now will affect which and how many people there will be in the future.

¹⁶ Presumably, on Barry's account, rules of justice should be neutral between every conceivable conception of the good. However, it is not immediately clear to me quite what this means or why one should think justice requires it.

¹⁷ This is not to suggest that any account of justice simply has to accept the motivations which people happen to have. I believe Barry right to be concerned with the question of the motivation to be just, but wrong in thinking that justice as impartiality happily connects with a motivation that is already widespread.

¹⁸ This is much more of a problem for Barry than for Scanlon. Since Scanlon purports to be offering a contractarian account of morality there are no moral considerations left outside his theory. However, Barry accepts that people may have a wide range of moral motivations and it is far less easy to see why the agreement motive should always out-weigh the other moral motives which people may have among their reasons for action.

¹⁹ In fact some non-Christians, for example, prefer a situation in which Sunday, a traditionally Christian day, is afforded a special place through legislation to a situation in which no religious day is afforded a special place. This is because they regard the latter as tantamount to favouring secularism.

²⁰ See especially his discussion of Gibbard in § 8.

²¹ Barry never takes seriously the possibility of a pluralist theory of justice. He presents the ideas of impartiality, reciprocity, and mutual advantage exclusively as competitors, without really considering whether each might give an adequate account of particular aspects, or limited sphere, of justice.

²² Again, there is a large issue here about the nature of a theory of justice; what can be expected from it and whether or not we can get along without the kind of theory that Barry, and so many other political philosophers, seem to think we need.

²³ In some cases, for reasons similar to those which Barry himself advances against the principle of preclusion as applied to abortion, people may find a situation more acceptable if it is seen as the outcome of imperfect political bargaining than if they are told it is a requirement of justice.

²⁴ What Barry needs to show is more than that a morally informed modus vivendi is potentially unstable. After all, in practice, we know that many people will not act as any theory of justice requires them to if they stand to gain considerably from violating such requirements. All principles of justice have a potential to be violated unless they can be shown to be always in everyone's interest, something neither Barry nor I would regard as remotely plausible.

²⁵ Barry rejects the view that ‘harm to others’ is a necessary condition for legislation, arguing that legislation may be permissible where others are adversely affected but cannot be said to be harmed, and that some paternalistic legislation need not be incompatible with justice as impartiality (see pp. 86–7).

²⁶ In fact Barry is curtly dismissive of this view claiming that ‘I have never … seen this [view] backed up by convincing evidence, and I do not believe it could be’ (p. 141). I should perhaps declare an interest here as I once tried to defend the position Barry rejects in ‘Toleration, Morality and Harm’, Aspects of Toleration, ed. Horton, J. and Mendus, S., London, 1985Google Scholar. My argument drew heavily on a similar view advanced in Phillips, D. Z. and Mounce, H. O., Moral Practices, London, 1969, esp. ch. 6Google Scholar. Maybe the view is untenable in such a strong form – although I continue to subscribe to it, I am not now quite so confident in some of the arguments as I once was – but it would be nice to see Barry take it seriously.

²⁷ I assume that what is important for justice as impartiality is that people agree, when not ill-informed or coerced, that something is bad. This is not of course to explain the reasons why people think something is bad, but for the purposes of the theory it is their agreement not their reasons (so long as they are not demonstrably mistaken) that matters.

²⁸ A good example of this problem is Barry's discussion of the compulsory wearing of seat belts and crash helmets (p. 87). Even after several readings it is quite unclear to me why legislation in this area is permissible but is neither required nor prohibited by justice as impartiality. If the harm involved in driving without a seat belt is agreed to be serious then presumably justice as impartiality should enjoin its prohibition. If, however, driving without a seat belt is a more or less harmless expression of a particular conception of the good then prohibitive legislation is unjust. But what is the reason for thinking such legislation neither just nor unjust, so long as the decision has been arrived at in a way which is procedurally correct? My own speculation is that Barry is implicitly relying on a notion of ‘real interests’ in this as in many of his arguments.

²⁹ How would justice as impartiality deal with the institution of duelling? This, historically at least, has formed an important expression of honour associated with a particular way of life. So long as people are not coerced into a duel I cannot see that justice as impartiality would have any reason to prohibit it.

³⁰ This possibility is briefly alluded to by Barry in his discussion of Charles Larmore but he does not take it very seriously (p. 167).

³¹ I am not thinking here primarily of circumstances of ‘fecklessness’ but, for example, of the fact that some people knowingly choose for fun to engage in activities such as mountain-climbing and deep-sea diving which involve very considerable risks to their health. Why does justice require that everyone should bear the costs of such choices when things go badly wrong? Why must it be unreasonable to expect people who engage in these activities to pay for at least part of the costs of their treatment (perhaps through medical insurance)?

³² An earlier draft of this paper was presented at New College, Oxford. I am very grateful for the discussion it received on that occasion, and to Cecile Fabré and Liz Frazer for their invitation. I have also benefited from the comments of Margaret Canovan, Andy Dobson, and Paul Kelly.